Postprint
This is the accepted version of a paper published in Journal of Evolutionary Biology. This paper has been peer-reviewed but does not include the final publisher proof-corrections or journal pagination.
Citation for the original published paper (version of record):
Grieshop, K., Stångberg, J., Martinossi-Allibert, I., Arnqvist, G., Berger, D. (2016)
Strong sexual selection in males against a mutation load that reduces offspring production in seed beetles.
Journal of Evolutionary Biology, 29(6): 1201-1210 https://doi.org/10.1111/jeb.12862
Access to the published version may require subscription.
N.B. When citing this work, cite the original published paper.
Permanent link to this version:
http://urn.kb.se/resolve?urn=urn:nbn:se:uu:diva-304536
Strong sexual selection in males against a mutation load that reduces offspring production in 1
seed beetles 2
3 4 5
Karl Grieshop1†, Josefine Stångberg1†, Ivain Martinossi1, Göran Arnqvist1 and David Berger1 6
7 1Department of Ecology and Genetics, Animal Ecology, Uppsala University, Norbyvägen 18D, 8
752 36 Uppsala, Sweden 9
10 † Both authors contributed equally to the study 11
12
13 Running title: Sexual and natural selection on new mutations 14
15 Key words: adaptation, genetic correlation, intralocus sexual conflict, pleiotropy, population 16
viability, sexual antagonism, sexual selection 17
18 Correspondence:
19
Karl Grieshop 20
Zooekologen, EBC 21 Uppsala Universitet 22 SE 752 36, Uppsala 23 Fax +46 (0)70 023 65 76
24 Email: karlgrieshop@gmail.com 25
Abstract:
26
Theory predicts that sexual reproduction can increase population viability relative to 27
asexual reproduction by allowing sexual selection in males to remove deleterious mutations 28
from the population without large demographic costs. This requires that selection acts more 29
strongly in males than females and that mutations affecting male reproductive success have 30
pleiotropic effects on population productivity, but empirical support for these assumptions is 31
mixed. We used the seed beetle Callosobruchus maculatus to implement a three-generation 32
breeding design where we induced mutations via ionizing radiation (IR) in the F0 generation, 33
measured mutational effects (relative to non-irradiated controls) on mating-pair productivity in 34
the F1, and effects on sex-specific competitive lifetime reproductive success (LRS) in the F2. 35
Regardless of whether mutations were induced via F0 males or females, they had strong 36
negative effects on male LRS, but a non-significant influence on female LRS, suggesting that 37
selection is more efficient in removing deleterious alleles in males. Moreover, mutations had 38
seemingly shared effects on mating-pair productivity and competitive reproductive success in 39
both sexes. Thus, our results lend support to the hypothesis that strong sexual selection on 40
males can act to remove the mutation load on population viability, thereby offering a benefit to 41
sexual reproduction.
42
Introduction 43
Sexual selection can act as a purifying force removing alleles with deleterious effects on 44
population mean fitness if the mutations that render individuals less successful in competition 45
over fertilization are also those that detriment offspring production (Zahavi 1975, Rowe & Houle 46
1996, Tomkins et al. 2004). This mutational pleiotropy can allow sexual selection to, at least 47
partly, compensate for the two-fold cost of sexual reproduction (Whitlock & Agrawal 2009). By 48
acting more strongly in males than females, sexual selection can remove inferior males of low 49
genetic quality from the mating pool, thereby reducing the population’s mutation load without 50
discernable demographic costs (Manning 1984, Agrawal 2001, Siller 2001, Lorch et al. 2003).
51
Whereas studies in Drosophila indicate that selection against new mutations is stronger in 52
males, little is known about such sex-biases in selection intensities in other organisms (reviewed 53
in: Whitlock & Agrawal 2009).
54
If mutations instead have sex-limited, or even opposing (i.e. sexually antagonistic), 55
fitness effects in the sexes, sexual selection on males would be inefficient at reducing mutation 56
load and could even increase the frequency of mutations that reduce female fecundity, 57
imposing a severe gender load on the population (Brooks 2000, Rice & Chippindale 2001, 58
Pischedda & Chippindale 2006, Arnqvist & Tuda 2010). The expected impact of sexual selection 59
on adaptive rates is therefore highly contingent upon the fitness effects of allelic variation at 60
loci experiencing sexually concordant versus sexually antagonistic selection (Bonduriansky &
61
Chenoweth 2009, Whitlock & Agrawal 2009). Recent theoretical approximations (e.g. Connallon 62
et al. 2010, Stewart et al. 2010, Connallon & Clark 2014) and empirical estimates based on 63
standing genetic variation in laboratory (e.g. Rice & Chippindale 2001, Fedorka & Mousseau 64
2004, Pischedda & Chippindale 2006, Bilde et al. 2009, Berger et al. 2014a) and wild populations 65
(e.g. Brommer et al. 2007, Foerster et al. 2007, Mainguy et al. 2009, Svensson et al. 2009, Tarka 66
et al. 2014, Barson et al. 2015) alike, suggest that natural populations harbor variable, but 67
potentially abundant, amounts of sexually antagonistic genetic variance for fitness. In 68
accordance, effects of sexual selection on rates of adaptation from standing genetic variation 69
are idiosyncratic and inconclusive (reviewed in: Candolin & Heuschele 2008, Whitlock & Agrawal 70
2009, Pennell & Morrow 2013).
71
Furthermore, mutations with sexually concordant fitness effects should be efficiently 72
eliminated (or fixed) by selection, while those with sexually antagonistic effects may not be 73
(Kidwell et al. 1977, Connallon & Clark 2012). Thus, allelic variation at sexually antagonistic loci 74
should contribute disproportionately to standing genetic variation for fitness (Long et al. 2012, 75
Connallon & Clark 2012; 2014, Berger et al. 2014b). Inferences based on standing genetic 76
variation, therefore, likely underestimate the potential for sexual selection to purge the genome 77
of novel deleterious mutations. Methods inducing de novo mutations may therefore be more 78
informative regarding the capacity for sexual selection to purge a population’s mutation load.
79
As mentioned above, several studies in Drosophila support the notion that selection 80
against new mutations is stronger in adult males than females (e.g. Sharp & Agrawal 2008, 81
MacLellan et al. 2009, Mallet et al. 2011; 2012, Clark et al. 2012, Sharp & Agrawal 2013).
82
However, sexual selection is surprisingly inconsistent across studies and mutations in its effect 83
on population level fitness, reported as being positive (e.g. Hollis et al. 2009), ineffectual (e.g.
84
McGuigan et al. 2011, Arbuthnott & Rundle 2012), or even negative (e.g. Hollis & Houle 2011, 85
Arbuthnott & Rundle 2012). Thus, while the sexes may share much of their developmental 86
genes, sexual selection in the adult stage could mostly target male-limited genes (see: Rice &
87
Chippindale 2001), weakening the potential for strong purifying sexual selection to remove 88
alleles with pleiotropic effects on female fecundity and juvenile survival.
89
Here, we measured the strength of sex-specific selection on novel mutations, and their 90
shared effect on population productivity and competitive adult reproductive success, in another 91
model organism, the seed beetle Callosobruchus maculatus. We induced a mutation load by 92
exposing individuals to ionizing (gamma) radiation (IR) and subsequently implemented a Middle 93
Class Neighborhood (MCN) breeding design (Shabalina et al. 1997) to minimize selection on the 94
induced mutations, allowing them to be passed through three subsequent experimental 95
generations.
96
To estimate the strength of selection on induced mutations, we compared competitive 97
lifetime reproductive success (LRS) of F2 adults originating from irradiated grandparents relative 98
to that of F2 controls originating from non-irradiated grandparents. The estimated strength of 99
selection was then compared across the sexes. Finally, we estimated the shared effect of 100
mutations on population productivity (measured in F1 adults) and male competitive LRS 101
(measured in F2 adults) by correlating family means of the two measures across generations.
102
Our results show that selection operates against new mutations in adult males, and that these 103
induced mutations had shared effects on male LRS and population productivity.
104 105 106
Methods 107
Study System 108
C. maculatus (Coleoptera: Bruchidae) is a pest of leguminous crops that has colonized most of 109
the tropical and subtropical regions of the world (Southgate 1979). Males and females are 110
sexually mature upon adult eclosion, and exhibit a polyandrous mating system (Miyatake &
111
Matsumura 2004). The eggs are glued onto the surface of dry beans and hatched larvae bore 112
into the beans, where they complete their life cycle.
113
The study population was isolated from Vigna unguiculata seed pods collected at a 114
small-scale agricultural field close to Lomé, Togo (06°10'N 01°13'E) during October and 115
November, 2010. Isofemale lines were created by mating a single male and female emerging 116
from the collected seeds. After establishment, isofemale lines were expanded to a population 117
size of approximately 200-300 adults and then kept on ca. 600 V. unguiculata seeds at 29o C, 118
55% RH and a 12L:12D photoperiod. They were cultured under this regime for ~30 generations 119
prior to the start of this experiment (see further: Berger et al. 2014b). Four isofemale lines were 120
randomly selected (from the 41 available for use) as the focal genetic backgrounds on which we 121
either induced mutations (in the case of treated beetles) or did not (in the case of controls). In 122
addition, a mixture of all the 41 isofemale lines was set up to create a reference population, 123
initiated 6 generations prior to the start of the experiment, against which focal individuals from 124
our experiment competed in the assays of competitive LRS (see below).
125 126
Inducing Mutations in the F0 Generation 127
We induced mutations using ionizing (gamma) radiation (IR) from a Cs137 source. IR causes 128
double strand breaks (DSB) to DNA, which occur naturally during recombination, and can 129
produce point mutations and deletions as a consequence of mistakes during their repair (Evans 130
& DeMarini 1999, Sudprasert et al. 2006, Shrivastav et al. 2008, Shee 2013). Importantly, the 131
number of DSB induced by IR is remarkably constant from bacteria to humans (ca.
132
0.005/Gy/Mbp: Daly 2012). This predictability has allowed the use of IR to induce mutation 133
loads and infer selection in a range of insect study systems (e.g. bulb mites: Radwan 2004, 134
Drosophila: Agrawal & Wang 2008, Maklakov et al. 2013, dung beetles: Almbro & Simmons 135
2014, seed beetles: Power & Holman 2015).
136
A pilot study was conducted to generate dose-response curves for F0 productivity (i.e.
137
the number of offspring produced by the irradiated individuals) upon sex-specific exposure to IR 138
(see electronic supplementary information, Fig. S1). These dose-response curves indicated that 139
20Gy was a suitable dosage for this experiment, inducing a quantifiable mutation load while still 140
allowing irradiated individuals to produce enough F1 offspring with which to conduct 141
experiments.
142
Egg-laden V. unguiculata seeds from each of the four isofemale lines were isolated in 143
order to collect virgin adults as they emerged. Zero-day-old virgins from each isofemale line 144
were separated by sex and held in 90mm ∅ petri dishes (n≈20 per container) and then assigned 145
randomly to one of four treatment categories: female-irradiated, male-irradiated, female- 146
control and male-control (Fig. 1). Males and females assigned to the male- and female- 147
irradiated categories, respectively, were exposed to 20 Gy of IR; whereas males and females 148
assigned to the male- and female-control categories, respectively, were not exposed IR, but 149
were otherwise treated exactly the same in terms of collection, handling, and holding container 150
density (Fig. 1). Two hours following the irradiation treatment the individuals from each of these 151
four treatment categories were paired with a zero-day-old virgin individual of the opposite sex 152
from their respective isofemale line in a petri dish (90mm ∅) containing ca. 100 V. unguiculata 153
seeds (Fig. 1). The pairs were kept together for their entire lifetime under the same abiotic 154
conditions stated above. The number of F1 offspring emerging from each F0 pair was counted;
155
this formed our measure of F0 productivity, which was used only to generate the dose response 156
curves (see above and Fig. S1). This procedure was repeated over two consecutive days, 157
generating two different cohorts from which families were derived—this structure was 158
maintained over generations throughout the experiment, and cohort was included as a fixed 159
effect when analysing the results (see Statistical Analysis). In total we set up 4-6 F0 pairs per 160
treatment category and genetic background.
161 162
F1 Productivity 163
From each F0 pair we created two F1 pairs by pairing randomly selected virgin male and female 164
offspring, generating a total of 8-12 F1 pairs per treatment category and genetic background.
165
(Fig. 1). This middle-class neighborhood (MCN) breeding design prevents selection from 166
operating on all but the unconditionally lethal mutations by allowing high- and low-fitness 167
individuals to contribute an equal number of offspring (in this case four) to the next generation 168
(Shabalina et al. 1997; Morrow et al. 2008). This was important as we aimed to measure and 169
relate the effects of mutations (induced in the F0) in the F1 and F2 generations, and therefore 170
could not allow selection to remove induced mutations over generations. The mating pairs were 171
kept under the same conditions stated above, and the F2 offspring that emerged from these F1
172
pairs were counted to estimate each F1 pair’s productivity, and used to assay male and female 173
LRS in the F2 generation (see further below) (Fig. 1).
174
We chose to construct the F1 pairs from within-family mating pairs (i.e. via full-sib 175
mating). This way, our breeding design preserved mutations induced in F0 such that F1 and F2
176
individuals from irradiated treatments had, on average, half of their genome exposed to IR, and 177
F1 and F2 individuals from the same family were more likely to share mutations induced in their 178
F0 relatives. Consequently, individuals were inbred one additional generation beyond the one 179
generation of inbreeding inherent in the establishment of the genetic backgrounds (isofemale 180
lines). While this detail of our breeding design may have lowered statistical power by rendering 181
a subsample of individuals homozygous for induced recessive mutations, increasing within-pair 182
variance for F1 productivity and F2 competitive LRS, this increase in the proportion of F2
183
homozygotes also increased the likelihood of detecting mutational effects. We note that the 184
productivity of the inbred F2 control individuals were not lower than what is usually observed 185
for this species in our lab, consistent with C. maculatus being resistant to multiple generations 186
of inbreeding (e.g. Tran & Credland 1995). Thus, this extra generation of inbreeding is in itself 187
unlikely to have affected our results.
188 189
F2 Competitive Lifetime Reproductive Success 190
Two randomly selected virgin F2 males and females from each F1 pair were used for estimating 191
each F1 pair’s male and female F2 competitive LRS (Fig. 1). Competitive LRS assays consisted of a 192
single focal individual placed in a petri dish (90mm ∅) containing ad libitum V. unguiculata 193
seeds together with a sterile virgin standard competitor of the same sex from the reference 194
population and two opposite-sex individuals from the reference population (a 1:1 sex ratio; Fig.
195
1). Competitor individuals were sterilized with a 100 Gy dose of IR, which, in the case of males, 196
still allows their sperm to function and fertilize eggs, but their zygotes die; this is standard 197
protocol among insects for revealing competitive fertilization success (Simmons 2001), which 198
we have successfully employed previously to reveal variation in competitive LRS (Berger et al.
199
2014b). The fertilized eggs of females receiving a 100 Gy dose of IR do not hatch (I. Martinossi, 200
unpublished data). Thus, both male and female competitive LRS assays included mating 201
competition, male assays also included sperm competition, and female assays included 202
competition for available oviposition sites. Since these assays represent an environment that 203
these beetles experience naturally in grain storage facilities (Southgate 1979, Fox 1993), they 204
also incorporate naturally occurring selection pressures, including but not limited to mate 205
searching, female mating resistance, competition over matings, sexual conflict over remating 206
rate, and female competition for oviposition sites. At the same time, these assays exclude 207
potentially ecologically relevant factors such as predation, adult food resources, and 208
fluctuations in population size and adult sex ratio. However, some of these aspects are likely 209
excluded from the natural habitat of these beetles as well (e.g. adult food availability is very low 210
on arid crop fields as well as in grain storage facilities). These assays were placed in the same 211
abiotic conditions stated above, where individuals competed for matings/fertilizations and laid 212
eggs for their entire lifespan. The number of individuals emerging from these assays was 213
counted to estimate sex-specific F2 competitive LRS (Fig. 1).
214 215
Statistical Analysis 216
All analyses were conducted in R v.3.2.3 (R core team 2014). Productivity and competitive LRS 217
were analyzed using Maximum Likelihood (ML) estimation in generalized linear mixed effects 218
models with a Poisson error structure and log-link function, implemented in the lme4 package 219
V. 1.1-10 (Bates et al. 2015). When analyzing productivity, fixed effects included treatment (i.e.
220
irradiated vs. control), sex-treated (i.e. male vs. female), and their interaction (Fig. 1). Genetic 221
background (i.e. isofemale line) crossed by treatment and sex irradiated were included as 222
random effects, assuring the correct level of replication for the main effects. We also blocked 223
out possible differences between cohorts by adding it as a main effect. These same terms were 224
used in a model with a binomial error structure to analyze the difference in the number of 225
males and females emerging from productivity assays—testing for sex differences in juvenile 226
survival. When modelling competitive LRS, we included sex-assayed (i.e. male or female LRS) as 227
an additional fixed effect crossed with treatment and sex-treated. Genetic background crossed 228
by treatment, sex-treated and sex-assayed, were included as random effects.
229
In the models on productivity and LRS we included each observation as a random effect 230
(i.e. “observation-level random effects"). This estimates the true residual variance in the model 231
rather than setting it equal to the mean of the response (which is only true for a perfectly 232
Poisson distributed variable) and thus accounts for overdispersion in the hypothesis testing 233
(Crawley 2012), providing a more conservative analysis. Statistical significance was evaluated by 234
likelihood ratio tests of models with and without the effects of interest using type-II sums of 235
squares in the car package V. 2.1-1 (Fox & Weisberg 2011).
236
To estimate selection coefficients along with their 95% credible intervals, we ran 237
Bayesian Markov Chain Monte Carlo simulations using the MCMCglmm package V. 2.22 for R 238
(Hadfield 2010) on data where the response variable (offspring produced) had been 239
standardized for each genetic background and sex by dividing all observations by the mean 240
number of offspring produced by each respective groups’ controls. Thus, the selection 241
coefficients were calculated as: s = 1-LRSIRR/LRSCON (i.e., in terms of relative fitness), and we 242
calculated credible intervals and P-values for selection coefficients (i.e. we tested if they were 243
significantly different from 0) in males and females based on the resampled Bayesian posterior 244
estimates. Except for using relative fitness as a normally distributed response variable the 245
model was identical to the one specified for the ML estimation using lme4. We used weak (nu = 246
10-6)gamma priors for our random effects where the variances were set as [total variance in 247
data / number of variance components] for each random effect term. Simulations started with a 248
burn-in phase (100,000 iterations) followed by 1,000,000 iterations during which posterior 249
estimates were sampled. The models ran with large sampling intervals (thin = 500) to minimize 250
autocorrelation (r < 0.05 for all parameters) of the stored posterior estimates. This generated an 251
effective sample size of 2000 uncorrelated posteriors of male and female selection coefficients 252
against the induced mutations (see Fig. 2a). In addition, we also ran models for each genetic 253
background and sex independently (i.e. in 8 separate models) to estimate sex-specific selection 254
coefficients on each genetic background (see Fig. 2b).
255
Finally, we calculated means for each F1 pair’s male and female competitive LRS 256
(measured in the F2) to estimate their (Pearson’s) correlation coefficients with productivity 257
(measured in the F1). To minimize the effect of standing genetic variation on the correlations we 258
blocked out main effects of genetic background. Thus, if there is positive mutational pleiotropy 259
between population productivity and male competitive LRS, we expect more positive 260
correlations across families in the irradiated treatments (carrying mutations with variable fitness 261
effects) relative to families of the control treatments.
262
263 264
Results 265
F1 Productivity 266
Offspring of irradiated parents had significantly lower productivity than controls overall (χ2 = 267
7.41, df = 1, p = 0.0065). However, the effect of treatment was clearly detectable via irradiated 268
fathers, but not mothers, as shown by a significant interaction between treatment and sex 269
irradiated (χ2 = 4.09, df= 1, p = 0.043) (Fig. S2; Table S1). There was no overall significant sex 270
difference in mutational effects on juvenile survival (χ2 = 0.98, df = 1, p = 0.322; Table S2).
271 272
F2 Competitive LRS: Sex-Specific Strengths of Selection on Induced Mutations 273
Overall, male and female individuals of irradiated grandparents had significantly lower 274
competitive LRS compared to control individuals (χ2 = 4.99, df = 1, p = 0.026). There was, 275
however, a tendency for an interaction between treatment and sex-assayed (χ2 = 2.71, df = 1, p 276
= 0.0997). Investigating this further by analyzing the sexes separately showed that male LRS was 277
strongly decreased by novel mutations (χ2 = 8.43, df = 1, p = 0.0037) while this effect was much 278
weaker and non-significant in females (χ2 = 2.38, df = 1, p = 0.123). These effects were 279
independent of the (grandparental) sex-treated, as indicated by a non-significant interaction 280
between treatment and sex-treated (full summary: Fig. S3, Table S3).
281
The Bayesian MCMC posterior estimates of selection coefficients (s) corroborated the 282
results from the analyses based on ML. Selection on the induced mutations was consistently 283
stronger in males relative to females both across sex-treated categories (Fig. 2a) and genetic 284
backgrounds (Fig. 2b). Again, there was no statistically significant sex difference in the strength 285
of selection (sM – sF = 0.10, CI: -0.03-0.26, PMCMC = 0.15), but selection was overall significant and 286
strong in males (sM = 0.20, CI: 0.04; 0.32, PMCMC = 0.010), whereas it was weak and non- 287
significant in females (sF = 0.07, CI: -0.01; 0.14, PMCMC = 0.08).
288 289
Correlations Between F1 Productivity and F2 Competitive LRS 290
Within the irradiated treatment, pooled over sex-treated categories, productivity was positively 291
correlated to competitive LRS of both females (r = 0.34, n = 80, p = 0.002) and males (r = 0.26, n 292
= 74, p = 0.024; Fig. 3). This was not the case among control individuals (with regard to male or 293
female LRS: r = 0.10, n = 82, p = 0.39 and r = 0.02, n = 87, p = 0.84, respectively), indicating that 294
novel mutations had shared effects on competitive LRS and productivity. There were no 295
significant differences in correlations depending on which sex was irradiated (Table S4).
296 297
298
Discussion 299
This study aimed to assess whether sexual selection can, at a relatively small demographic cost, 300
act to remove mutations that are detrimental to population mean fitness. For this to be the 301
case, mutations must firstly be selected more strongly in males than females, and secondly 302
detriment both male reproductive success and overall population productivity. We found i) that 303
induced mutations had strong fitness effects in adult males but not adult females, and ii) a 304
positive correlation between male reproductive success and productivity in irradiated 305
treatments, but not in control treatments, indicating that novel mutations may generally have 306
shared effects on male reproductive success and population productivity in seed beetles. Taken 307
together, our results thus offer support for the theoretical prediction that sexual selection in 308
males can offer an evolutionary benefit to sexual reproduction by reducing mutation load at a 309
small demographic cost (Manning 1984, Agrawal 2001, Siller 2001).
310
We induced mutations either via males or females in the F0 generation, and in both 311
cases point estimates of selection against the mutations were greater in males (Fig. 2a). Thus, 312
potential male bias in the strength of sexual selection against new mutations seems unlikely to 313
be attributed to mutations induced on the unprotected hemizygous Y-chromosome. Positive 314
mutational pleiotropy between male fitness and population productivity can alone compensate 315
for the two-fold cost of reproducing sexually if the intensity of selection on males is greater than 316
on females and the genome-wide deleterious mutation rate is sufficiently high (Agrawal 2001, 317
Siller 2001). Indeed, despite the overall strength of selection against novel mutations varying 318
across genetic backgrounds, point estimates of selection coefficients were consistently two to 319
three times greater in males relative to females within each genetic background (Fig. 2b).
320
Importantly, since our assays measured effects on adult competitive LRS, they do not 321
give a complete picture of the sex-bias in selection acting across the entire life cycle. For 322
example, including ecological factors and life stages that invoke the same intensity of selection 323
in males and females could reduce the overall sex-bias in selection against a novel mutation 324
with male-biased effects on competitive LRS. Indeed, our analysis of juvenile survival indicated 325
no significant difference in selection between the sexes (Table S2). Additionally, other ecological 326
aspects of these beetles that were not included in our selection estimates, such as more 327
extensive mate search in males and host search in females, could affect sex differences in 328
selection against novel mutations.
329
Previous studies investigating the effect of sexual selection on adaptation have reached 330
mixed results (reviewed in Whitlock & Agrawal 2009), which likely reflects the wide variety of 331
techniques, mating systems and evolutionary histories of the experimental populations studied.
332
Recent examples highlight some of this complexity. For example, Lumley et al. (2015) subjected 333
treatments of flour beetles to ~50 generations of experimental evolution at different intensities 334
of sexual selection, and then subjected replicated lineages from these treatments to single-pair 335
full-sib inbreeding. Lineages from populations evolving under intense sexual selection on males 336
tolerated sustained inbreeding for a greater number of generations relative to those from 337
populations evolving under enforced monogamy or intense sexual selection on females.
338
Tolerance to inbreeding is indicative of the level of mutation load (Charlesworth & Charlesworth 339
1999, Charlesworth & Willis 2009). Thus, Lumley et al. (2015) demonstrated that enhanced 340
sexual selection on males reduced populations’ accumulating mutation load.
341
In contrast, Chenoweth et al. (2015) studied the fixation of single nucleotide 342
polymorphisms (SNPs) across populations maintained over 13 generations under experimental 343
evolution treatments varying in the strength of both natural and sexual selection. While as 344
many as 80 SNPs showed statistically significant differences among the selection treatments, 345
only 6 SNPs showed aligned responses across the sexual selection and natural selection 346
treatment. Moreover, for 43 of the 80 SNPs, the effect of sexual selection when applied 347
simultaneously with natural selection, was to oppose the response observed when natural 348
selection was applied in isolation. This last result implies that sexual selection impeded 349
adaptation and the authors provided additional evidence showing that males directed courtship 350
and harassment disproportionally towards high quality females (a form of interlocus sexual 351
conflict), thereby offering a relative benefit to smaller females with lower fecundity (Chenoweth 352
et al. 2015).
353
The discrepancy between these two recent landmark studies may serve to illustrate the 354
opposing outcomes of sexual selection that can be expected when selection is either allowed to 355
act over longer periods of time to target ongoing mutational input like in the study of Lumley et 356
al. (2015), or when it acts on standing genetic variation over shorter periods of time like in the 357
study of Chenoweth et al. (2015), for which purifying selection has already ensued, and the 358
remaining sexually antagonistic genetic variation in combination with interlocus sexual conflict 359
is likely to swamp the beneficial effects of purifying sexual selection (Whitlock & Agrawal 2009).
360
Turning the focus to two recent studies that employed similar methods to ours, Power 361
and Hollman (2015), found results that they interpret as opposite to ours despite using the 362
same system (C. maculatus). Using (X-ray) IR, they created mutated populations with 363
significantly reduced egg-to-adult survivorship, but no difference in the number of offspring 364
produced, relative to control populations. Then, looking within their mutated populations only, 365
they compared females that had been mated via enforced monogamy to females that were 366
mated by the winner of three competing males (allowing pre-copulatory sexual selection).
367
Perhaps understandably, they found that females produced the same number of offspring 368
regardless of whether or not pre-copulatory sexual selection was allowed. They conclude that 369
sexual selection did not benefit female productivity, but their results are difficult to interpret 370
considering the dosage of IR they used did not elicit a reduction in female productivity, relative 371
to controls, from the start, and considering that pre-copulatory sexual selection is typically weak 372
relative to post-copulatory sexual selection in this species (Fox et al. 2007, Fritzsche & Arnqvist 373
2013).
374
In contrast, Almbro and Simmons (2014) recently argued that sexual selection was 375
effective at increasing population fitness by purging a mutation load induced by (gamma) IR in 376
the dung beetle Onthophagus taurus. However, the induced mutations had no discernible 377
effects on female fecundity and only affected the measured male traits. Not surprisingly, the 378
implemented sexual selection treatment improved some of the male performance traits in the 379
following generations, but had no measurable effect on how the induced mutation load 380
affected female fecundity, suggesting pronounced sex-specificity of mutational effects.
381
The significant positive correlation between male reproductive success and productivity 382
we report here is consistent with the induced mutations having shared effects on these two 383
measures in our seed beetle population. The fact that this correlation was ≈0 in the control 384
treatment, as well as in the base population from which the four genetic backgrounds originate 385
(Berger et al. 2016, in revision), further reiterates the difference in sex-specificity of fitness 386
effects expected for novel mutations versus standing genetic variation.
387
Nevertheless, two points deserve specific consideration. First, when estimated over 388
multiple mutations induced across the entire genome, the correlation between male 389
reproductive success and population productivity provides a quantitative estimate of the 390
directionality of mutational effects on the two variables averaged over all mutations. In our 391
study, this correlation ranged between 0.21 (males irradiated) and 0.34 (females irradiated), 392
indicating that far from all mutations had shared effects on the two variables. Since our 393
estimates of F1 pair means from which we calculated correlations were based on low sample 394
sizes, measurement error is likely to have caused our correlations to fall below unity, and this 395
measurement error is likely to have been further exaggerated by F2 individuals being either 396
heterozygous or homozygous for the induced mutations (see Methods). However, this is 397
unlikely to fully explain the low correlations because the corresponding correlations between 398
productivity and female reproductive success for both male- and female-irradiated categories 399
were, as expected, stronger (0.29 and 0.42, respectively; Table S4). This implies that sexual 400
selection on males has the potential to purge only a fraction of those mutations with negative 401
effects on population productivity in C. maculatus. Indeed, in the extreme case, the underlying 402
reason for observing stronger selection in males could be due to sexual selection acting with 403
particular efficacy on those mutations with largely male-limited effects, which would greatly 404
reduce the population-level benefits of sexual selection. Characterizing selection intensities on 405
alleles with sex-limited versus sexually concordant fitness effects therefore remains an 406
important challenge for understanding the role of sexual selection in promoting population 407
mean fitness, which has only just begun with the study of selection on single mutations in 408
isolation in Drosophila (see Introduction).
409
Second, since we induced mutations in lineages kept isolated throughout the three 410
generations of the experiment, it is possible that a positive correlation between F1 productivity 411
and F2 reproductive success may have been generated by variation among families in the 412
number of mutations rather than variation in the effect sizes of mutations with shared effects 413
on the two traits, a caveat that applies generally to studies inducing mutation loads to study 414
sexual selection (Whitlock & Agrawal 2009), as well as to those that study trait and intersexual 415
correlations across mutation accumulation lines. The two alternative explanations are not 416
mutually exclusive and we cannot rule out that this second mechanism may be partly 417
responsible for the observed positive correlation. If so, however, it would imply that our F0
418
individuals varied substantially in their ability to repair DNA damage within each genetic 419
background, since the number of DSB in cells exposed to a given dosage of a given type of IR 420
appears to be fixed (Daly 2012), and we blocked out overall differences among genetic 421
backgrounds when estimating correlations.
422
One final detail of our study design worth addressing is that our F1 productivity measures 423
were significantly lower than controls when it was F0 males that were irradiated, but not when 424
F0 females were irradiated (Fig. 3 and Fig. S2). This could indicate a lower threshold for the 425
number of mutations tolerated/passed on by female gametes relative to male gametes (in line 426
with the sex differences in response to our 20 Gy dosage, Fig. S1), such that more detrimental 427
mutations were filtered out in the F0 generation when coming in through females, whereas 428
more detrimental mutations coming in through males were filtered out in the F1 generation.
429
Nevertheless, our F2 LRS estimates did not differ significantly between sex-treated categories 430
(i.e. did not seem to depend on whether or not males’ Y-chromosomes were mutated), 431
rendering this detail of our findings inconsequential to our interpretations.
432
In summary, we have provided empirical support for the hypothesis that sexual selection 433
has the potential to remove mutations that reduce population viability at a low demographic 434
cost, by generating strong selection in males against mutations with shared effects on male 435
reproductive success and population productivity. This finding is congruent with theoretical 436
expectations and contributes to a growing body of literature aiming to evaluate the ability of 437
sexual selection to counterbalance the two-fold cost of sex across a wide variety of study 438
organisms.
439 440 441
Acknowledgements:
442
This research was supported by the European Research Council (GENCON AdG-294333; to GA), 443
the Swedish Research Council (621-2010-5266, to GA), and the Stiftelsen för Zoologisk Forskning 444
(to KG). We thank Johanna Liljestrand Rönn for help with experiments. We thank Bo Stenerlöw 445
for access to the irradiation facilities. The data supporting this research have been uploaded to 446
the Dryad data repository (accession # xxx).
447
References 448
Agrawal, A.F. 2001. Sexual selection and the maintenance of sexual reproduction. Nature, 411:
449
692-695.
450
Agrawal, A.F. & Wang, A.D. 2008. Increased transmission of mutations by low-condition 451
females: evidence for condition-dependent DNA repair. PLoS Biology, 6: 389-395.
452
Almbro, M. & Simmons, L.W. 2014. Sexual selection can remove an experimentally induced 453
mutation load. Evolution, 68: 295-300.
454
Andersson, M. 1994. Sexual selection. Princeton University Press, Princeton, New Jersey.
455
Arbuthnott, D. & Rundle, H.D. 2012. Sexual selection is ineffectual or inhibits the purging of 456
deleterious mutations in Drosophila melanogaster. Evolution, 66: 2127–2137.
457
Arnqvist, G. & Tuda, M. 2010. Sexual conflict and the gender load: correlated evolution between 458
population fitness and sexual dimorphism in seed beetles. Proc. R. Soc. B Biol. Sci., 277: 1345- 459
1352.
460
Barson, N. J., Aykanat, T., Hindar, K., Baranski, M., Bolstad, G. H., Fiske, P., ... & Kent, M. (2015).
461
Sex-dependent dominance at a single locus maintains variation in age at maturity in 462
salmon. Nature.
463
Bates, D., Maechler, M., Bolker, B., Walker S. 2015. Fitting Linear Mixed-Effects Models Using 464
lme4. Journal of Statistical Software 67: 1-48.
465
Berger, D., Berg, E.C., Widegren, W., Arnqvist, G. & Maklakov, A.A. 2014a. Multivariate 466
intralocus sexual conflict in seed beetles. Evolution, 68: 3457-3469.
467
Berger, D., Grieshop, K., Lind, M.I., Goenaga, J., Arnqvist, A. & Maklakov, A.A. 2014b. Intralocus 468
sexual conflict and environmental stress. Evolution, 68: 2184-2196.
469
Bonduriansky, R. & Chenoweth, S.F. 2009. Intralocus sexual conflict. Trends Ecol. Evol., 24: 280- 470
288.
471
Brooks, R. 2000. Negative genetic correlation between male sexual attractiveness and survival.
472
Nature, 406: 67-70.
473
Brommer, J.E., Kirkpatrick, M., Qvarnström, A. & Gustafsson, L. 2007. The intersexual genetic 474
correlation for lifetime fitness in the wild and its implications for sexual selection. PLoS One, 8:
475
e744.
476
Candolin, U. & Heuschele, J. 2008. Is sexual selection beneficial during adaptation to 477
environmental change? Trends Ecol. Evol., 23: 446–452.
478
Charlesworth, B. & Charlesworth, D. 1999. The genetic basis of inbreeding depression. Genet.
479
Res., 74: 329–340.
480
Charlesworth, D. & Willis, J.H. 2009 The genetics of inbreeding depression. Nat. Rev. Genet. 10:
481
783-796.
482
Chenoweth, S.F., Appleton, N.C., Allen, S.L. & Rundle, H.D. 2015. Genomic evidence that sexual 483
selection impedes adaptation to a novel environment. Current Biology, 25: 1860-1866.
484
Clark, S.C.A., Sharp, N.P., Rowe, L. & Agrawal, A.F. 2012. Relative effectiveness of mating success 485
and sperm competition at eliminating deleterious mutations in Drosophila melanogaster. PLoS 486
One, 7: e37351.
487
Connallon, T., Cox, R.M., & Calsbeek, R. 2010. Fitness consequences of sex-specific selection.
488
Evolution, 64: 1671-1682.
489
Connallon, T. & Clark, A.G. 2012. A general population genetic framework for antagonistic 490
selection that accounts for demography and recurrent mutation. Genetics, 190: 1477-1489.
491
Connallon, T. & Clark, A.G. 2014. Balancing selection in species with separate sexes: Insights 492
from Fisher’s geometric model. Genetics, 197: 991-1006.
493
Crawley, M. J. (2012). The R book. John Wiley & Sons.
494
Daly, M. J. 2012. Death by protein damage in irradiated cells. DNA repair, 11: 12-21.
495
Evans, H.H. & DeMarini, D.M. 1999. Ionizing radiation-induced mutagenesis: radiation studies in 496
Neurospora predictive for results in mammalian cells. Mutat. Res., 437: 135–150.
497
Fedorka, K.M. & Mousseau, T.A. 2004. Female mating bias results in conflicting sex-specific 498
offspring fitness. Nature, 429: 65-67.
499
Foerster, K., Coulson, T., Sheldon, B.C., Pemberton, J.M., Clutton-Brock, T.H., Kruuk, L.E.B. 2007.
500
Sexually antagonistic variation for fitness in Red deer. Nature, 447, 1107-1109.
501
Fox, C.W. 1993. A quantitative genetic analysis of oviposition preference and larval performance 502
on two hosts in the bruchid beetle, Callosobruchus maculatus. Evolution, 47: 166–175.
503
Fox, C.W., Stillwell, R.C., Moya-Larano, J. 2007. Variation in selection, phenotypic plasticity, and 504
the ecology of sexual size dimorphism in two seed-feeding beetles. In: Sex, size, and gender 505
roles: evolutionary studies of sexual size dimorphism. (D.J. Fairbairn, W.U. Blanckenhorn & T.
506
Székely, eds), pp. 88-96. Oxford University Press, New York, New York.
507
Fox J, Weisberg S. 2011. An (R) Companion to Applied Regression, Second Edition. Thousand 508
Oaks CA: Sage. URL: http//socserv.socsci.mcmaster.ca/jfox/Books/Companion.
509
Fritzsche, K. & Arnqvist, G. 2013. Homage to Bateman: sex roles predict sex differences in sexual 510
selection. Evolution, 67: 1926-1936.
511
Hadfield, J.D. 2010. MCMC methods for multi-response generalized linear mixed models: the 512
MCMCglmm R package. Journal of Statistical Software, 33: 1-22.
513
Hollis, B. & Houle, D. 2011. Populations with elevated mutation load do not benefit from the 514
operation of sexual selection. J. Evol. Biol., 24: 1918–1926.
515
Hollis, B., Fierst, J. L. & Houle, D. 2009. Sexual selection accelerates the elimination of a 516
deleterious mutant in Drosophila melanogaster. Evolution, 63: 324–333.
517
Kidwell, J.F., Clegg, M.T., Stewart, F.M. & Prout, T. 1977. Regions of stable equilibria for models 518
of differential selection in two sexes under random mating. Genetics, 85: 171-183.
519
Long, T.A.F., Agrawal, A.F., Rowe, L. 2012. The effect of sexual selection on offspring fitness 520
depends on the nature of genetic variation. Curr. Biol., 22: 204-208.
521
Lorch, P.D., Proulx, S.R., Rowe, L. & Day, T. 2003. Condition-dependent sexual selection can 522
accelerate adaptation. Evol. Ecol. Res., 5: 869–881.
523
Lumley, A.J., Michalczyk, L., Kitson, J.J.N., Spurgin, L.G., Morrison, C.A., Godwin, J.L., Dickinson, 524
M.E., Martin, O.Y., Emerson, B.C., Chapman, T. & Gage M.J.G. 2015. Sexual selection protects 525
against extinction. Nature, 7557: 470-473.
526
MacLellan, K., Whitlock, M.C., & Rundle, H.D. 2009. Sexual selection against deleterious 527
mutations via variable male search success. Biol. Let., 5: 795–797.
528
Mainguy J., Côté, S.D., Festa-Bianchet, M. & Coltman D.W. 2009. Father-offspring phenotypic 529
correlations suggest intralocus sexual conflict for a fitness-linked trait in a wild sexually 530
dimorphic mammal. Proc. R. Soc. B Biol. Sci., 276: 4067-4075.
531
Maklakov, A.A., Immler, S., Løvlie, H., Flis, I. & Friberg, U. 2013. The effect of sexual harassment 532
on lethal mutation rate in female Drosophila melanogaster. Proc. R. Soc. B Biol. Sci., 280:
533
20121874.
534
Mallet, M., Bouchard, J., Kimber, C. & and Chippindale, A.K. 2011. Experimental mutation- 535
accumulation on the X chromosome of Drosophila melanogaster reveals stronger selection on 536
males than females. BMC Evol. Biol., 11: 156.
537
Mallet, M.A., Kimber, C.M. & Chippindale, A.K. 2012. Susceptibility of the male fitness 538
phenotype to spontaneous mutation. Biol. Let., 8: 426–429.
539
Manning, J.T. 1984. Males and the advantage of sex. J. Theor. Biol., 108: 215-220.
540
McGuigan, K., Petfield, D. & Blows, M.W.. 2011. Reducing mutation load through sexual 541
selection on males. Evolution, 65: 2816–2829.
542
Miyatake, T. & Matsumura, F. 2004. Intra-specific variation in female remating in 543
Callosobruchus chinensis and C. maculatus. J. Insect Physiol. 50: 403-408.
544
Morrow, E.H., Stewart, A.D. & Rice W.R. 2008. Assessing the extent of genome-wide intralocus 545
sexual conflict via experimentally enforced gender-limited selection. J. Evol. Biol., 21: 1046–
546
1054.
547
Pennell, T.M. & Morrow, E.H. 2013. Two sexes, one genome: the evolutionary dynamics of 548
intralocus sexual conflict. Ecology and Evolution, 3: 1819-1834.
549
Pischedda, A. & Chippendale, A.K. 2006. Intralocus sexual conflict diminishes the benefits of 550
sexual selection. PloS Biol, 4: e356.
551
Power, D.J. & Holman, L. 2015. Assessing the alignment of sexual and natural selection using 552
radiomutagenized seed beetles. J. Evol. Biol., 28: 1039-1048.
553
R Core Team. 2014. R: A language and environment for statistical computing. R foundation for 554
statistical computai, Vienna, Austria. URL http://www.R-project.org/
555
Radwan, J. 2004. Effectiveness of sexual selection in removing mutations induced with 556
ionization. Ecol. Let., 7: 1149–1154.
557
Rice, W.R & Chippindale, A.K. 2001. Intersexual ontogenetic conflict. J. Evol. Biol., 14: 685-693.
558
Rowe, L., & Houle, D. 1996. The lek paradox and the capture of genetic variance by condition 559
dependent traits. Proc. R. Soc. B Biol. Sci., 263: 1415–1421.
560
Shabalina, S.A., Yampolsky, L.Y. & Kondrashov, A.S. 1997. Rapid decline of fitness in panmictic 561
populations of Drosophila melanogaster maintained under relaxed natural selection. PNAS, 94:
562
13034–13039.
563
Sharp, N.P. & Agrawal A.F. 2008. Mating density and the strength of sexual selection against 564
deleterious alleles in Drosophila melanogaster. Evolution, 62: 857–867.
565
Sharp, N.P. & Agrawal A.F. 2013. Male-biased fitness effects of spontaneous mutations in 566
Drosophila melanogaster. Evolution, 67: 1189–1195.
567
Shee, C., Hastings, P.J. & Rosenberg S.M. 2013. Mutagenesis associated with repair of DNA 568
double-strand breaks under stress. In: Stress-Induced Mutagenesis (D. Mittelman, ed) pp. 21-40.
569
Springer, New York, New York.
570
Shrivastav, M., De Haro, L.P. & Nickoloff, J.A. 2008. Regulation of DNA double-strand break 571
repair pathway choice. Cell research, 18: 134-147.
572
Siller, S. 2001. Sexual selection and the maintenance of sex. Nature, 411: 689–692.
573
Simmons, L.W. 2001. Sperm Competition and Its Evolutionary Consequences in the Insects.
574 575
Princeton University Press, Princeton, NJ.
576 577
Southgate, B.J. 1979. Biology of the Bruchidae. Ann. Rev. Entom. 24: 449-473.
578
Stewart, A.D., Pischedda, A. & Rice, W. 2010. Resolving intralocus sexual conflict: genetic 579
mechanisms and time frame. Journal of Heredity, 101: S94-S99.
580
Sudprasert, W., Navasumrit, P. & Ruchirawat, M. 2006. Effects of low-dose gamma radiation on 581
DNA damage, chromosomal aberration and expression of repair genes in human blood cells. Int.
582
J. Hyg. Environ. Health, 209: 503–511.
583
Svensson, E.I., McAdam, A.G. & Sinervo, B. 2009. Intralocus sexual conflict over immune 584
defense, gender load, and sex-specific signaling in a natural lizard population. Evolution, 63:
585
3124-3135.
586
Tarka, M., Åkesson, M., Hasselquist, D., Hansson, B. 2014. Intralocus sexual conflict over wing 587
length in a wild migratory bird. Am. Nat., 183: 62-73.
588
Tomkins, J.L., Radwan, J., Kotiaho, J.S. & Tregenza, T. 2004. Genic capture and resolving the lek 589
paradox. Trends Ecol. Evol., 19: 323–328.
590
Tran, B. & Credland, P.F. 1995. Consequences of inbreeding for the cowpea seed beetle, 591
Callosobruchus maculatus (F.)(Coleoptera: Bruchidae). Biol. J. Linn. Soc., 56: 483-503.
592
Whitlock, M. C. & Agrawal, A.F. 2009. Purging the genome with sexual selection: reducing 593
mutation load through selection on males. Evolution, 63: 569–582.
594
Zahavi, A. 1975. Mate selection – a selection for a handicap. J. Theor. Biol., 53: 205-214.
595 596
Figure captions:
597 598
Figure 1: Methodological schematic followed for each of 4 genetic backgrounds. Each treatment 599
(irradiated or control) contained male and female ‘sex-treated’ categories. F0 individuals 600
indicated by a lightning bolt had their whole genomes exposed to 20 Gy of IR (indicated by IR 601
symbols). They passed half their genomes to their F1 offspring (indicated by half IR symbols). F1
602
pairs from the same F0 parents produced F2 offspring (the number of which was each F1 pairs’
603
productivity) with half their genomes consisting of grandparental DNA exposed to IR (also 604
indicated by half IR symbol). F2 individuals were used to estimate each F1 family’s sex-specific 605
competitive LRS. Parentheticals indicate the number of replicate pairs for each treatment and 606
sex-treated category of each genetic background in the F0, for each F0 pair in the F1, and for 607
each F1 pair in the F2. 608
609
Figure 2: Bayesian estimates (posterior modes ± 95% credible intervals) of selection coefficients 610
against genome wide induced mutations in males and females of C. maculatus. Selection on 611
new mutations tended to be stronger in males relative to females, depicted a) across the two 612
sex-treated categories in which either male or female grandparents were irradiated, and b) for 613
each of the four genetic backgrounds pooled across sex-treated categories.
614 615
Figure 3: Family-level correlation between F1 family productivity and F2 male competitive LRS.
616
Confidence ellipses depict the bivariate distributions (mean ± 95% CI). Families formed by 617
control males and females are pooled for clarity and depicted by the hatched ellipse and white 618
circle (mean = 1). Families in which F0 females were irradiated are depicted by the grey ellipse 619
and triangle, and families in which F0 males were irradiated are depicted by the black ellipse and 620
circle.
621
